OpenAI is reportedly partnering with Qualcomm and MediaTek to design a custom smartphone processor targeting 300 400 million annual shipments by 2028, with Luxshare Precision as the exclusive manufacturing partner. The chip is architected around AI as operating system principles,…
OpenAI has 800 million weekly users and no operating system. That is the pressure point driving a reported deal that sounds like a paradox: the company building AI so capable it could make the iPhone obsolete is working with the Taiwanese manufacturer that already builds a significant share of Apple's iPhones.
According to TF International Securities analyst Ming-Chi Kuo, OpenAI is working with Qualcomm and MediaTek to design a custom smartphone processor, with mass production targeted for 2028 and annual shipment ambitions of 300 to 400 million units. Luxshare Precision, the contract manufacturer that already assembles a substantial portion of Apple's iPhones under Apple's direction, is slotted as the exclusive system co-design and manufacturing partner. Luxshare has spent years trying to challenge Foxconn's dominant position in Apple's supply chain. OpenAI just handed it the most significant prize available: building the device that could make the iPhone irrelevant, using a contract manufacturer that Apple itself depends on.
The architecture Kuo describes is not an app with AI features. It is AI as the operating system itself, with simpler tasks running on-device and more complex inference offloaded to the cloud. The processor design priorities are power consumption, memory hierarchy management, and small-model execution. Sam Altman, who a day earlier posted on X that it is "a good time to seriously rethink how operating systems and user interfaces are designed," has given the clearest signal yet that OpenAI's hardware ambitions extend beyond the AI earbuds and pen-shaped device it has been rumored to be building.
The 300-to-400-million-unit target is ambition, not a forecast. No new smartphone entrant has displaced Apple or Samsung at meaningful volume in over a decade. The incumbents have supply chain leverage, carrier relationships, and developer ecosystems that new entrants cannot replicate quickly. The chip specifications and supplier decisions are not expected to be finalized until late 2026 or early 2027, according to Kuo. Mass production, if the timeline holds, is three years away.
That timeline is the skeptic's strongest argument. OpenAI announced a partnership with Jony Ive's io Products studio in early 2026, a $6.5 billion acquisition that positioned it as a hardware company overnight, and has not yet shipped a consumer device, as Wccftech reported. It announced a custom AI accelerator partnership with Broadcom in October 2025 targeting 10 gigawatts of deployments, but that infrastructure is still being built.
The deeper question is what happens to the app economy if OpenAI succeeds. Apple generates over $20 billion in quarterly revenue from the App Store, per industry estimates. The iPhone ecosystem is worth hundreds of billions annually to developers, carriers, accessory makers, and Apple itself. That ecosystem exists because users navigate apps to get things done. An AI agent that completes tasks without opening an app does not need an app store, iOS, or Google Play. If the smartphone survives as an app-delivery platform, it survives on Apple's terms. If it does not, the question becomes who captures the value instead.
Qualcomm and MediaTek have historically been component suppliers operating under the direction of device OEMs and platform companies like Apple and Google. If this partnership cements them as co-architects of an AI-first device category rather than passive vendors, they ascend from the commodity layer to the platform layer. They have every incentive to make it work.
OpenAI has not confirmed the plans. Qualcomm and MediaTek have not confirmed the plans. Luxshare declined to comment. The report comes from a single analyst sourcing supply chain contacts, credible but not unimpeachable. If all three parties deny the report, the evidentiary floor drops significantly. What is clear is why the structural logic is sound: OpenAI has grown to over 800 million weekly active users, and now has a plausible path to owning the full stack from silicon to interface. Whether it can execute before Apple and Google respond is the question that will define the next three years of the smartphone market.